Archive for May, 2009

HDHomerun and Windows 7 Media Center

The HDHomerun is an ethernet based dual television tuner. It’s compatible with Linux, Mac and Windows and since I have a mixed OS household, thought it would be the way to go.

On the Silicondust forum, I’ve noticed a number of posters who reported problems in the Windows 7 release candidate. Although I’ve only had the Homerun on my network for about 4 days, so far it’s been entirely straightforward and easy to use.

It was literally:

  • Unpack box
  • Plug in cable
  • Plug in power
  • Install drivers
  • Use Media Center to scan for TV channels
  • Watch TV

Since my whole house aside from the little hub around my cable modem is wireless, the picture quality, especially on HD broadcasts can be a bit choppy when I’m “doing other stuff” on the Internet and trying to watch/record TV at the same time. But when the wireless network is quiescent, the picture quality is great.

Originally, I had dismissed the idea of a MythTV box “just” to record television, but I’m thinking now it might be a good idea.  I could record the TV shows on the Myth server, and copy the files to my Home Server, and then from the Home Server, all the PCs and Macs in my house could watch the shows. Probably not a good idea to toss this idea to the wife right now, but it’s in the back of my head. :)

No Comments

First impressions of Windows 7 RC

I’ve been running Windows 7 RC on two of my home systems for about a week now.

System 1 is a Lenovo S10 Netbook with an N270 ATOM processor, 1 GB of RAM and an 80GB disk slice. (Dual-boots OpenSolaris B111; see this post for more about that.) The netbook is where I installed Windows 7 beta back in February.

System 2 is a self-built PC with a Q8400 quad-core Intel processor, 8 GB RAM, 1TB of disk, Nvidia 8800 GTX (with the G93 GPU) and a TrendNet Wireless N USB adapter.  It was previously running Vista SP1 Ultimate x64 and was built around November 2008. (This system used to have a Linksys PCI Wireless N interface, but I had problems with it under Vista and Windows 7. It was solid under XP SP3, for whatever that’s worth.)

I did a Vista->W7 upgrade of System #2, and a clean install on System #1. The upgrade process was very smooth. I was a tiny bit frustrated by the upgrade advisor though.  It was great that it pointed out incompatible Windows 7 software (daemon-tools) that I had on Vista, but it was annoying because it only identitfied one show-stopper at a time. So everytime I corrected an issue and restarted the upgrade I felt like I was gambling on whether I’d pass the upgrade advisor check.  It would be better if the installer identified all of the potential conflicts on the same screen, so at least it’s clear that when those problems are resolved you can immediately start an upgrade.  The upgrade itself took about an hour and when it was finished I was amazed that all of my old software was seemingly automagically reinstalled.  I’ve had absolutely zero problems with the software and since Nvidia is on the ball with WQHL-certified Windows 7 video drivers, I was able to start several 3D gaming engines quickly.  The Windows Experience Index on this system was rated at 5.9 (out of a possible 7.9) the lowest performing component being the hard drive.

With the netbook, I did a clean install on that system since Microsoft clearly expressed the desire that W7 Beta installs get fresh installations instead of in-place upgrades. This time, I installed Windows 7 32 bit using a 16GB USB stick, following the directions here.  Since this was a clean install and from flash memory it took much less time than the DVD-based system #2 upgrade. After the upgrade was complete, the system tested out with a WEI score of 2.1, with the CPU being the weakest system component. I was a bit surprised the netbook scroed so low, because according to Microsoft guidelines, a system should have a minimum WEI of 3 to run Aero. The native Intel 945 graphics controller seems to do a fine job with Aero, however.  The one area where sluggishness is especially noticable on the netbook is on websites under IE8 with a lot of fancy javascript (like Yahoo Mail, for example.) That isn’t really a Windows 7 issue, per se, but more a function of IE8′s rather slow Javascript engine. And although I have Firefox installed on both computers, I’ve been trying to be a good tester and use IE8 for my every day browsing.

Other impressions of W7: The desktop artwork themes that come “out of the box” are stunning.  I have the landscape photo theme on System #2 and the “United States” photo theme on system #1. Both of them are really really “Oooh” factor type photos. I love that Gadgets are no longer forced along the side of display 1. I have dual LCD panels on system #2 and although you can undock gadgets in Vista, the sidebar app itself still has an overlay on the right side of display 1. Performance wise, I’ve noticed that system shutdown and start up are noticably faster than Vista – much more in line with XP.  On system #2 I have a CPU/Memory gauge gadget running all the time and I’ve noticed that memory usage is 10-15% better and CPU usage, even during peak use is about 5-20% lower.

All in all?  This is the best version of Windows I’ve used. Yep, even better than XP which looks really long-in-the-tooth compared to Aero. W7 looks amazing, functions well and so far has been very stable on all of my applications.  I will definitely upgrade all of my PCs to W7 RTM once its available.

3 Comments

Building OpenSolaris USB installer from Windows

The “official” way to build a USB install of OpenSolaris is to run usbcopy from a command prompt inside a running OpenSolaris system.  Well, ok, that’s not very convenient for me, so I was so happy to see Hiroshi Chonan’s Live USB creator for Windows.

Grab the USB image from genunix.org and you’re about 97% complete with this task.

I wanted to post here because a) this is extremely useful and cool, and b) it’s not explicitly stated on Hiroshi’s webpage, but if you’re running the USB creator on Vista or Windows 7, you need to right-click the program file and choose “Run as administrator” to make sure that the boot sector is appropriately installed on your USB stick.

As you can see in my next post, I’ve installed Windows 7 and OpenSolaris 2009.06 preview (B111a) on my Lenovo S10 netbook using USB sticks for both operating systems.

1 Comment

Wireless support in OpenSolaris 2009.06 for Lenovo S10 netbook

I’ve spent a ridonculous amount of time trying to get my Lenovo S10 netbook’s Broadcom wireless chipset to work in OpenSolaris 2009.06 (build 111, actually.)

I wanted to record here the recipe I used to make this work, since working directions are spread out all over teh Interwebs.

First off, if you’re using the LiveCD or LiveUSB images from genunix.org to install your OpenSolaris system, you’ll need to find a wired connection and jack in to install the following filesets:

  1. SUNWgcc
  2. SUNWhea
  3. SUNWflexlex
  4. SUNWgm4

This is accomplished by doing something like:

$ pfexec pkg install SUNWgcc SUNWhea SUNWflexlex SUNWgm4

Next, download the current ndis code.  The most current as of this blog post is ndis-1.2.3, which you can download from http://www.opensolaris.org/os/community/laptop/wireless/ndis/ndis-1.2.3-prelim.tar.gz

Unpack the tarball:

$ tar zxvf ndis-1.2.3-prelim.tar.gz

Figure out what kind of Windows driver you need to download.  Instructions for doing this are pretty well covered on the ndis project page at OpenSolaris.org.

Download and unpack your Windows driver.  In OpenSolaris, you can simply right-click on the .exe and choose “uncompress” from the context menu in the file manager.  You’ll want to find and copy the appropriate Windows XP .sys and .inf drivers into ~/ndis-1.2.3/i386.

Convert the .inf file from Windows UTF-16 text into good old ASCII then copy of the driver.sys file to ndis.sys.

$ iconv -f utf-16 -t ascii bcmwl5.inf > ndis.inf
$ cp bcmwl5.sys ndis.sys

Next (and this was an endless source of frustration for me) make sure that Sun’s make program (/usr/ccs/bin/make) not GNU make (typically /usr/bin/make) processes the ndis Makefile by setting your shell’s PATH variable correctly:

$ export PATH=/usr/ccs/bin:$PATH

If you do not do this step, you will get strange error messages complaining about not finding header files like cdef.h and bunch of other heartache.

Finally, you’ll want to start building the ndis wrapper.

$ cd ~/ndis-1.2.3/i386
$ make ndiscvt
$ make ndis.h
$ make ndis
$ make ndisapi
$ pfexec cp bcmndis /kernel/drv/bcmndis
$ pfexec cp ndisapi /kernel/misc

Now we’re ready to load the ndis wrapped drivers into the kernel.  Important: The PCI id information below 14e4,4315 is for my particular hardware. You will have to put in your own PCI id information when you want to insert the driver.

$ pfexec add_drv -i '"14e4,4315"' bcmndis
$ pfexec ifconfig bcmndis0 plumb

The following things are broken in ndis-1.2.3 (patches welcome… :)

  • 64 bit kernel support (you have to boot into 32 bit kernel to use ndis)
  • WPA2 (WPA may or may not work – supplying my WPA2 password to nwam causes kernel panics on my netbook)

The good news is that nwam should pick up your new interface and help you get on to your wifi network and remember all of the relevant settings for that interface.

2 Comments

AT&T free wifi offer at Starbucks is useless

Prepare for a righteous rant.

I’m not sure who came up with the idea that I need to jump through 32 flaming hoops to access the purportedly “free” wireless at Starbucks locations provided by AT&T to iPhone users.

It’s a lot more accurate to say that the access is “without charge” rather than “free” because it takes me approximately 2 minutes to get to the front of the line at the Starbucks location closest to my work. It takes approximately 3 minutes to actually get my “free” wifi code from AT&T. Time is money. As I get older I really understand that more and more.

Net effect: no iPhone usage – or at best, I use the over-the-air Internet if I want to kill some waiting-in-line time.

Attention AT&T: Your “free” wifi offer stinks and so does your cellular network. I can only hope that rumors of Verizon getting its own iPhone are true.

What’s really depressing is that everyone expects a new iPhone model this summer and if I get it (my original iPhone 1.0 from June 2007 is coming off contract) then I have to sign up for another 2 years on AT&T. Guess how much I’m looking forward to that.

No Comments